Five Eyes warns AI models capable of major cyberattacks are months away

Five Eyes Warn Advanced AI Hacking Models Are Months Away

The cybersecurity agencies of the Five Eyes intelligence alliance issued a rare joint statement on Monday warning that frontier AI models capable of causing major cyber damage are mere months from public availability, urging organizations worldwide to act immediately.

The statement, released by agencies from the United States, United Kingdom, Canada, Australia, and New Zealand, cautioned that “frontier AI models are expected to surpass existing industry forecasts, significantly altering offensive and cyber capabilities,” adding that “the timeline is not years; it is merely months.”theguardian

Australia’s Cyber Security Centre published the statement, which warned that “cyber risk assumptions can become outdated in months, not years,” and called on both government and corporate leaders to bolster their defenses before the window closes.cyber

A New Generation of Cyber-Capable Models

The warning comes amid a turbulent period for frontier AI development. Anthropic launched its Fable 5 model on June 9 — described as a public-facing version of its restricted Mythos system — only for the U.S. Commerce Department to order it pulled three days later under an export control directive citing national security. The model had topped every major AI benchmark during its brief availability.thenextweb

OpenAI has also entered the cyber arena with Daybreak, a cybersecurity platform integrating its GPT-5.5 and Codex systems to automate vulnerability detection and patching for partners including Cisco and CrowdStrike.cyberscoop

Testing of Anthropic’s Mythos model by the NSA reportedly demonstrated the system’s ability to penetrate classified U.S. Cyber Command systems within hours during controlled evaluations, according to reports citing NSA Director General Joshua Rudd.news-pravda

Building on Months of Escalating Concern

Monday’s statement is the latest in a series of increasingly urgent Five Eyes advisories on AI risks. On May 1, the same agencies published “Careful Adoption of Agentic AI Services,” a 30-page guide — the first coordinated Five Eyes advisory targeting autonomous AI systems — that outlined 23 distinct risks and over 100 best practices. That document urged organizations to “assume that agentic AI systems may behave unexpectedly” and to prioritize “resilience, reversibility and risk containment over efficiency gains.”theregister

The Sydney Morning Herald reported that agency heads emphasized the shrinking window between vulnerability discovery and exploitation, stressing that resilience is now essential for “maintaining operational continuity and market confidence.”smh